Research On Indoor Localization Technology Based On Wireless Fidelity

With the development of wireless network and extensive application of pervasive computing technology, Location Based Service(LBS) has inspired a growing demand of the location information of mobile users. At present, the outdoor positioning system based on satellite signal is very mature, but it cannot be used in the complex indoor environment, so the indoor positioning technology has become a hot topic in the research field. Meanwhile, the rapid development of wireless network technology, due to its fast communication, convenient deployment as well as other characteristics, has enabled the WiFi-based indoor positioning technology to attract wide attention and research. Therefore, based on the location fingerprint positioning technology which is the most widely used among the WiFi-based indoor positioning methods, this thesis studies the corresponding improved algorithm and strategy focusing on the existing problems. Further, based on the improved algorithm and strategy, a set of indoor positioning system combining positioning and map has been realized on the Android platform. The specific research contents are as follows:First, in WiFi-based indoor positioning technology, the traditional Weighed K-nearest Neighbor(WKNN) algorithm based on Received Signal Strength(RSS) is not able to adaptively get valid Access Points(AP) as well as relatively high matching accuracy of reference points. Aiming at this problem, the WKNN algorithm with self-adaptive matching and preprocessing function is studied in this thesis. In this algorithm, the RSS values of APs are sorted according to the network state adaptively, and the top M APs coupled with the corresponding M APs in the reference points are selected to participate in the calculation of the matching in order to optimize the traditional fingerprint positioning algorithm. Experimental results show that the proposed algorithm can achieve more than 30% of the positioning error improvement which can effectively improve the positioning accuracy and positioning stability.Second, aiming at the problem of positioning accuracy degradation and stability reduction due to Received Signal Strength(RSS) detection variance of different devices that has caused the fingerprint database unable to directly match mobile devices of different brands and models, this thesis studies a compensation strategy for RSS detection difference based on Cosine Similarity(CS) which introduces cosine similarity as the metric to determine whether different devices could conduct compensation for RSS variances. For device pairs that meet the metric standard, compensation for RSS detection variances is made by using ratio correction method which has effectively solved the positioning accuracy degradation and stability reduction causing by RSS detection variances between different devices.Third, based on the Android platform, a WiFi indoor positioning system is realized which organically combines indoor positioning technology and indoor map technology. At offline training stage, within just a click on the map, the coordinates of the reference points and floor information can be obtained from the map database and be saved to a fingerprint database, which greatly simplifies the offline training process. Meanwhile, each reference point could be visually displayed on the map, making the reference point training more simple, intuitive and reasonable. Online positioning stage can also use the map data to make positioning more accurate, intuitive, and efficient. On this basis, the system realizes the functions of single point positioning, mobile positioning and historical trajectory queries and so on, and provides an interface for future expansion of the system. The last, results of the actual measurements and verifications in this system show that the above two algorithms proposed in this thesis are effective.